Elastic's Agent Builder is generally available, Elastic Workflows in tech preview

Elastic said its Agent Builder is generally available as the company aims to bring context engineering into developer workflows. Elastic also launched Elastic Workflows, an automation engine built into the Elasticsearch platform to execute actions across systems.

Agent Builder runs on Elastic Cloud Serverless and aims to differentiate from the market by leveraging Elastic's core strengths. Elastic is looking to give enterprises the ability to build agents quickly via chat with relevance and customization. Elastic Workflows, which is in tech preview, is an automation engine that handles rules-based tasks and unified automation and operational data directly within Elasticsearch. 

Elastic’s duo—Elastic Agent Builder and Elastic Workfows—is designed to address input reliability for models and AI agents. Agent Builder provides the context and reasoning, while Workflows provides the orchestration and execution. Ken Exner, chief product officer at Elastic, said Agent Builder and Workflows are designed to give teams “a single system that delivers both intelligent reasoning and dependable automation.”

The company launched Agent Builder in October in private preview. Speaking during an investment conference in December, Elastic CEO Ashutosh Kulkarni said "the idea here is you start to build agents directly on top of your data because the context is what gives LLMs meaning in what they need to do. And that's the secret sauce for most businesses."

• Welcome to the context chorus: There’s no AI without context

Elastic's Agent Builder is built directly into Elasticsearch and integrates chat, retrieval, tools, workflows, security, and observability into one surface. Those parts of the agent lifecycle typically require vector databases, retrieval augmented generation (RAG) pipelines, query translators and orchestration often cobbled together. Custom agents can use Elastic search, analytics and security tools. 

This integration between Elastic Agent Builder, Elasticsearch and Elastic Workflows are part of a market shift toward context workflows and management, argued Constellation Research analyst Michael Ni. Why? One-time retrieval doesn't scale well enough for AI agents. 

• Agentic Automation Doesn’t Fail at Reasoning. It Fails at Inputs.

"With its Agent Builder launch, Elastic positions Elasticsearch as a context-engineering foundation bringing together hybrid retrieval with security controls and developer workflows to help teams build context-driven agents on their data. Elastic’s Agent Builder then illustrates the need," said Ni. "It’s not about making agents smarter; it’s about giving them execution context that is assembled, governed, and observable at runtime to ensure its agents, or any agents leveraging the platform, can operate safely in production."

Ni added that the architecture underneath agentic AI is changing as semantic layers and catalogs move closer to runtime, multi-step workflows and governance converge and observability shifts from offline reporting to ongoing feedback loops. 

Elastic's Agent Builder includes the following:

• A full agent-building system with custom tools, personas, execution control and observability.
• An execution surface that executes tools and workflows inside the same platform that stores and ranks data.
• The ability to combine multiple types of searches (lexical, sparse semantic, dense vector and reranking) with inference.
• Security that's enforced during execution instead of after retrieval. 
• Interoperability and open standards with Model Context Protocol (MCP), API integration and LLM choices. 

Elastic Agent Builder enables developers to chat with data, reason when a request is unclear and invoke tools to gather context from systems. Agent Builder is built to work with Elastic Workflows, which provides business-grade guardrails, reusable actions and tools for gathering information and taking action.